Wolfgang Denk wd@denx.de wrote on 2011/01/18 22:25:04:
Dear Joakim Tjernlund,
In message OFBBE91305.B19350D9-ONC125781C.007192FC-C125781C.007421BB@transmode.se you wrote:
Ah, finally you make sense to me. I actually tested this with mainline on my board so it is not completely untested in mainline.
As your board itself is not in mainline this _is_ completely untested for mainline.
It's a nice and appreciated RFC patch or even example implementation, but I fail to see arguments why we should add this to mainline.
Well, you have to start somewhere and as this involves asm changes in start.S it would be pretty dangerous add these without being able to test. The idea is that once some version of this patch is in, interested parties can apply the same concept on their boards too.
Such testing can be done anywhere, in some test branch. I don't think mainline is the right place for such an intrusive and experiemental feature.
Do you have such a branch where you can apply this?
Finally, I would like to remind you about [PATCH] PowerPC: Move -fPIC flag to common place
I have not seen any test reports on this, so I hesitate to apply it (the move to a common place is no problem, of course, but I'm not sure about changing -fPIC into -fpic). I think I remember problems with this in the past; there have even been commits to "use '-fPIC' _instead_ of '-mrelocatable'" in the past. I don't remember the details, but I'd like to see some independent testing before this goes in.
Right, this can be confusing. Without -mrelocatable you don't get fixups so this has to stay as fixups is a must nowadays.
Assuming you don't need fixups, -fpic needs what went into the linker scripts a little while ago so -fpic had little chance before that.
[PATCH] PowerPC: Add support for -msingle-pic-base
Same here. This hits a large number of boards, but I have seen zero test reports.
Scott and Kim, can you give these 2 patches a spin?